The coronavirus pandemic slowed clinical studies, particularly in oncology. [1] But it also sped up acceptance of electronic data collection, or eSource, a concept that has been discussed in the oncology research community for years. Now, as researchers and hospitals restart clinical trials, they’re discovering that newly embraced eSource-related innovations have inaugurated a system of generating evidence that promises to increase the efficiency and quality of oncology trials while lowering cost.

The rise of eSource is powering three trends in oncology trials today:

  • Real-world data (RWD) is streamlining clinical trial design and helping to digitally identify patients who can participate in studies.
  • eConsent forms and new thinking about consent and longer-term patient engagement are driving new research approaches.
  • Researchers are realizing that digitally confederated clinical networks could improve upon the current heterogeneous clinical systems that silo data from imaging, molecular testing, practice management, and laboratory information management systems.

Applying RWD as studies are conceived rather than at their conclusions will save researchers time and money. AI can comb through large datasets to identify study cohorts, defining such factors as inclusion and exclusion criteria and treatment arms. Most importantly, investigators can seek out and test study approaches early, avoiding dead ends.

Along with identifying patient cohorts, AI can be used to match individual patients to studies [2]. There’s a lot of room for improvement in this area: Only five percent of cancer patients participate in trials, although many more are eligible [3]. AI today is leveraging existing structured clinical, claims and molecular data to infer what’s missing. As AI continues to advance in healthcare with natural language processing [4] – it will aid in extracting and understanding the patient journey from unstructured data such as handwritten physician notes or scanned interpretation from a diagnostic lab attached to patient’s electronic medical records, (EMR). Combined together, data and AI will streamline patient identification even more. As AI is integrated into clinical workflows, it can simultaneously analyze standard treatment of care options and potential clinical trials for patients. This ability will be central to success.

Port Worthy
Novasep Top 6 qualities to look for in a CDMO partner for ADCs
TRP Targeted Radiopharmaceuticals Summit
American Lung Association

Changes taking place in the consent process are also generating fresh thinking that will enable new research approaches. The eConsent approach is far more than just electronic versions of the paper forms that patients sign with pens in clinics. It also enables the multimedia explanation of clinical trials, as well as remote consent. Preliminary evidence suggests that eConsent could be leveraged for lifelong patient engagement [5]. The eConsent approach could make it easier for researchers to engage with patients over the entire life cycle of a therapy, to determine long-term outcomes, and to stay in touch for potential participation in future clinical research.

Advertisement #3

In later phases of clinical trials, the informed consent forms, EMR, and study data can digitally pre-populate the electronic case report forms (eCRF) in electronic data capture (EDC) systems used by biopharma sponsors and government agencies. As researchers transition to fully digital clinical trials based on eSources, the eCRFs will provide a seamless audit trail and generate libraries of clinical resources.

Over time, we believe that the border between patients’ clinical data, what goes into the EDCs during a trial, and what subsequently gets submitted to FDA, will become an integrated eSource system of evidence streamlining clinical research. Digital clinical trials will deliver datasets that can be analyzed by biopharma, contract research organizations, and regulators — without resorting to EDCs.

High levels of interoperability are paramount to the success of these developments, however. Data access will need to be complete, and data source standards will need to be written to consistently map and codify a variety of data sources. (See, for example, “Leveraging the CDISC Standards to Facilitate the Use of Electronic Source Data in Clinical Trials.”) Rules and models will be needed to take unstructured data and turn it into measurable variables proven to have come from peer-reviewed approaches or, in the case of novel approaches, supported by retrospective RWD analyses.

The third major trend that will shape the development of digital clinical trials in oncology will be the creation of more digitally confederated clinical networks. These will need to be defined at the outset of trials to ensure that they share the same software-as-a-service (SaaS) models, and to ensure that data from highly heterogeneous sources meets technical, informatic, and codification requirements to produce high-confidence datasets.

All of these efforts are needed to increase the productivity of oncology clinical trials, which one study found were successful less than four percent of the time [6]. While no one can wave a magic wand to suddenly digitize oncology clinical trials, creating the smooth, integrated end-to-end digital workflows that researchers desire will certainly help. The approach solves a number of small, difficult problems and breaks though many technological bottlenecks. Creating digital clinical trial networks will take patience, collaboration and commitment. But the end result will be the speedy development of new therapies, and better healthcare for cancer patients.

Patrycja Vasilyev Missiuro, PhD, is VP of Product and Judith Mueller, PhD, is VP of Data Science at ConcertAI.

[1] JAMA Oncol. 2021;7(3):458-460. doi:10.1001/jamaoncol.2020.7600.
[2] Woo, Marcus. Nature 573, S100-S102 (2019).
[3] Unger JM, Cook E, Tai E, Bleyer A. The Role of Clinical Trial Participation in Cancer Research: Barriers, Evidence, and Strategies. Am Soc Clin Oncol Educ Book. 2016;35:185-198. doi:10.1200/EDBK_156686
[4] Guergana K. Savova, Ioana Danciu, Folami Alamudun, Timothy Miller, Chen Lin, Danielle S. Bitterman, Georgia Tourassi and Jeremy L. Warner. Use of Natural Language Processing to Extract Clinical Cancer Phenotypes from Electronic Medical Records. Cancer Research November 1 2019 (79) (21) 5463-5470.
[5] Zeps N, Northcott N, Weekes L. Opportunities for eConsent to enhance consumer engagement in clinical trials. Med J Aust. 2020;213(6):260-262.e1. doi:10.5694/mja2.50732.
[6] Chi Heem Wong, Kien Wei Siah, Andrew W Lo, Estimation of clinical trial success rates and related parameters, Biostatistics, Volume 20, Issue 2, April 2019, Pages 273–286.

Featured image: Clinical Trials. Photo Courtesy: © 2016 – 2022 Foltolia/Adobe. Used with permission.

Advertisement #5